Comment 36 for bug 1596075

There is a big gap in the n2 neutron logs that is hiding the original requests from several retries. The two statements below are 24 seconds apart.

2016-07-11 02:47:41.528 248 ERROR oslo_db.sqlalchemy.exc_filters File "/usr/local/lib/python2.7/dist-packages/pymysql/cursors.py", line 161, in execute
2016-07-11 02:48:05.043 242 DEBUG neutron.quota.resource [req-9213bd6a-eb41-4c3e-9b03-6ff7daf3b668 a4aab8472ebb41faa98f89e1dd34858a 469300d96f9f406cacda5f001d387e52 - - -] Quota usage for subnet was recalculated. Used quota:18. count /usr/local/lib/python2.7/dist-packages/neutron/quota/resource.py:295

Can you try using the log_dir and log_file options for oslo log in case the stderr printing is being rate limited by something? I'm having difficulty finding a single failed request ID that can be tracked from its initial request all of the way to a response because there is so much lost log data.

An example is '172.16.170.0/24'. You can see in the pcap that it got an overlap error but there is nothing in the logs about receiving a create request for that payload or even that it generated an overlap error for that CIDR.